Learning Where Outcomes Change:Credit-Addressable Reasoning for Multimodal Geometry

2026-08-31

AI Summary

本文提出信用可寻址推理方法，通过Code-CoT和CE-GRPO技术解决多模态几何推理中视觉关系提取与多步演绎的问题，提高了推理准确率。

Abstract

Multimodal geometry reasoning requires VLMs to extract precise visual relations and preserve them through multi-step deduction. Existing free-form traces obscure the decisions that determine the answer, and trajectory-level reinforcement learning distributes a single terminal signal across the entire response. We introduce credit-addressable reasoning, in which the semantic units exposed during inference also define where learning compares alternatives and assigns credit. We instantiate this principle with Code-CoT, which retains the diagram, represents visual relations as line-addressable executable code, and organizes reasoning into typed events, and CE-GRPO, which selects event boundaries using structural priors and type-normalized entropy, samples complete continuations from shared prefixes, and converts outcome differences into localized advantages. Across nine geometry benchmarks, CE-GRPO achieves an average accuracy of 76.04, outperforming Qwen3-VL-8B and trajectory-level GRPO by $8.09$ and 3.43 points, respectively. Its relative advantage increases with the number of intermediate events, demonstrating the value of representation--optimization co-design for long, dependency-heavy multimodal reasoning.

Problem

• Multimodal Geometry
• Visual Relations
• Multi-step Deduction
• Credit-Addressable Reasoning

Innovation

• credit-addressable reasoning
• Code-CoT
• CE-GRPO
• multimodal geometry reasoning